Anomaly Detection in BACnet systems
Meeting 29
Title: Meeting with Dr. Boutros
Date: 12/04/2020
Time: 1 pm
Location: Zoom Conference Call
Attendees: Sofian Ghazali, Rahul Balamurugan, Zahid Kamil
Absences: None
Action Items: Make changes to project based on advice
Meeting Agenda:
-
Present project progress to Dr. Boutros
-
Consult on k-modes, silhouette selection of clusters, and dissimilarity measures
-
Consult next steps
-
Verify if testing methods are valid under the present circumstances
Discussion and Notes:
Planned since a few days ago, the meeting was originally to be held the previous evening but had to be pushed to its present time due to Dr. Boutros' schedule. The main point he commented on was the lack of equipment to test the model on, due to which he said our current method of testing by creating an anomalous data set were valid. He approved of us using the Jaccard score for k-modes as well.
Dr. Boutros recommended we focus on frequency anomaly detection as well and that we not change the program much, simply adding a separate module. Since the preprocessing was fairly simple given the data was numerical, we were all in consensus of following through by completing the module using k-means and a similar anomaly detection algorithm as the one we designed for message anomalies.
Next meeting action items:​
-
Practicing for second presentation
Meeting Adjournment: 2:20 pm
Meeting Minutes prepared by: Rahul Balamurugan